[PATCH v4 6/8] x86: Add hardware prefetch control support for x86

[tarumizu.kohei at fujitsu.com]

> As much fun as it is to see a function like this, that is not ok, and never guaranteed
> to keep working, sorry.  Please find the device properly, never assume a specific
> driver model topology as they are guaranteed to break over time and never
> supposed to be static like this.

I will consider a method that does not depend on a specific driver
model topology in the next version.

> How do you know attributes are to be marked read only after init?  Do other parts
> of the kernel do that?  I don't think the driver core guarantees that at all.

I will fix a proper marking in the next version.